Severity: high.
41% of CALLS (20 549 of 49 544) carry confidence < 0.5, via unique_name (0.38)
and suffix_match (0.01–0.28). These are name guesses, yet they share an edge type
with LSP-resolved calls, so any consumer that does not filter gets noise.
Measured precision of the unfiltered intra-service CALLS layer: 52% (26/50).
With confidence >= 0.9: 92% (46/50).
Visible in the top-called list: filter resolves into a test module, "".join and
os.path.join both resolve to a _FakeQuery.join test stub.
Ask: emit low-confidence guesses as CALL_REFERENCE — the edge type already
exists and is barely used (163 edges here) — or expose a threshold so that CALLS
carries resolved edges only. Consumers currently cannot tell the layers apart
without knowing to filter on an undocumented property.
Observed on v0.10.8 (build 996bad5fe6fb), macOS arm64, Python monorepo (~36k nodes / ~179k edges). Happy to provide a minimal reproduction repo if useful.
Severity: high.
41% of
CALLS(20 549 of 49 544) carryconfidence < 0.5, viaunique_name(0.38)and
suffix_match(0.01–0.28). These are name guesses, yet they share an edge typewith LSP-resolved calls, so any consumer that does not filter gets noise.
Measured precision of the unfiltered intra-service
CALLSlayer: 52% (26/50).With
confidence >= 0.9: 92% (46/50).Visible in the top-called list:
filterresolves into a test module,"".joinandos.path.joinboth resolve to a_FakeQuery.jointest stub.Ask: emit low-confidence guesses as
CALL_REFERENCE— the edge type alreadyexists and is barely used (163 edges here) — or expose a threshold so that
CALLScarries resolved edges only. Consumers currently cannot tell the layers apart
without knowing to filter on an undocumented property.
Observed on v0.10.8 (build
996bad5fe6fb), macOS arm64, Python monorepo (~36k nodes / ~179k edges). Happy to provide a minimal reproduction repo if useful.